oxixml-xml

A Pure-Rust XML 1.0/1.1 pull parser and writer that mirrors the quick-xml 0.41 public API (module paths, type names and method signatures) closely enough that most downstream code can switch over with a one-line dependency change. It provides the zero-copy Event model, a namespace-aware NsReader, an events-based Writer, escaping utilities, and — behind opt-in features — a serde (de)serializer/serializer, tokio async I/O, and non-UTF-8 decoding. Part of the OxiXML workspace, the COOLJAPAN Pure-Rust foundation for XML and RDF processing.

Install

[dependencies]
oxixml-xml = "0.1.2"

# with optional features:
# oxixml-xml = { version = "0.1.2", features = ["serialize", "encoding"] }

Usage

use oxixml_xml::escape::{escape, unescape};
use oxixml_xml::events::{BytesStart, Event};
use oxixml_xml::name::QName;

// Build a start tag and inspect its name and attributes.
let start = BytesStart::new("greeting").with_attributes([("lang", "en")]);
assert_eq!(start.name(), QName(b"greeting"));
let (key, value) = {
    let attr = start.attributes().next().expect("one attribute").expect("valid");
    (attr.key.as_ref().to_vec(), attr.value.to_vec())
};
assert_eq!((key.as_slice(), value.as_slice()), (&b"lang"[..], &b"en"[..]));

// Wrap it in an event; events borrow their payload bytes.
let event = Event::Start(start);
assert_eq!(&*event, b"greeting lang=\"en\"".as_ref());

// Escape and unescape round-trip.
assert_eq!(escape("a < b & c"), "a &lt; b &amp; c");
assert_eq!(unescape("a &lt; b &amp; c").expect("valid"), "a < b & c");

For reading a document event-by-event, use oxixml_xml::reader::Reader (or the namespace-aware NsReader); for writing, oxixml_xml::writer::Writer and its create_element builder. See the crate documentation for full examples.

Features

arbitrary (arbitrary::Arbitrary impls on the event types, for fuzz targets), async-tokio (async I/O via tokio), encoding (non-UTF-8 input, backed by the Pure-Rust oxixml-encoding crate instead of encoding_rs), escape-html (HTML5 named entity recognition), overlapped-lists and serialize (serde (de)serialization through the de/se modules), and serde-types (serde impls on a few event types, plus the serde_helpers module). All are off by default.

arbitrary mirrors quick-xml 0.41's optional arbitrary dependency exactly: BytesStart, BytesEnd, BytesText, BytesCData, BytesDecl, BytesPI, BytesRef and Event all implement arbitrary::Arbitrary, with the same validation quick-xml applies — BytesStart requires an all-alphanumeric, non-empty name; BytesText requires an all-alphanumeric string but allows it to be empty; every other type accepts any string. A fuzz harness can call Event::arbitrary directly on raw fuzzer bytes instead of hand-rolling a generator. Off by default, so the crate stays dependency-free (see Replaces) unless explicitly opted into.

Replaces

quick-xml — oxixml-xml is a Pure-Rust, unsafe-free replacement with the same 0.41-era API surface.

Conformance

Gated on the W3C XML Conformance Test Suite (xmlts), restricted to the subset applicable to a non-validating, non-external-entity-reading processor: 984/988 not-wf cases correctly rejected (4 cases are intentional exceptions, tracked in an allowlist) and 478/478 valid/invalid documents parsed successfully. XML 1.1 and Namespaces in XML 1.1 cases are included. See tests/xmlts.rs for the exact skip policy and allowlist rationale.

License

Licensed under the Apache License, Version 2.0.